Improving Quantized Model Performance in Qualitative Analysis with Multi-Pass Prompt Verification
The study focuses on enhancing the performance of quantized large language models (LLMs) in qualitative analysis. It introduces a multi-pass prompt verification method to reduce hallucinations and improve accuracy. The findings indicate that while lower-bit models face challenges, the proposed method stabilizes their performance for qualitative research.
- ▪Quantized LLMs are increasingly used for qualitative analysis due to their efficiency and lower resource requirements.
- ▪The study evaluates the impact of different quantization levels on the performance of LLaMA-3.1 using expert and non-expert responses.
- ▪The proposed multi-pass prompt verification method improves the stability and accuracy of low-resource LLMs.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Computation and Language arXiv:2605.20193 (cs) [Submitted on 4 Apr 2026] Title:Improving Quantized Model Performance in Qualitative Analysis with Multi-Pass Prompt Verification Authors:Aisvarya Adeseye, Jouni Isoaho, Adeyemi Adeseye View a PDF of the paper titled Improving Quantized Model Performance in Qualitative Analysis with Multi-Pass Prompt Verification, by Aisvarya Adeseye and 2 other authors View PDF HTML (experimental) Abstract:Quantized Large Language Models (LLMs) are used more often in qualitative analysis because they run fast and need fewer computing resources. This study examines how different lower bits quantization levels (8-bit, 4-bit, 3-bit, and 2-bit) and quantization types affect the performance of LLaMA-3.1 (8B) on qualitative analysis.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
